Skip to content

perf(semantic): resolve_references_for_current_scope without a temp Vec#22599

Merged
graphite-app[bot] merged 1 commit into
mainfrom
claude/perf-semantic-resolve-refs-no-tempvec
May 21, 2026
Merged

perf(semantic): resolve_references_for_current_scope without a temp Vec#22599
graphite-app[bot] merged 1 commit into
mainfrom
claude/perf-semantic-resolve-refs-no-tempvec

Conversation

@Dunqing

@Dunqing Dunqing commented May 19, 2026

Copy link
Copy Markdown
Member

Summary

SemanticBuilder::resolve_references_for_current_scope is the early-resolution hook fired at function / arrow / catch boundaries — it resolves the references collected since a checkpoint against the outer scope chain so that e.g. function f(x = outerRef) {} correctly binds outerRef to the outer scope rather than the function body.

The old implementation copied the pending slice into a fresh Vec on every call:

let refs = self.unresolved_references.slice_from(checkpoint).to_vec();
self.unresolved_references.truncate(checkpoint);
for (name, reference_id) in refs { ... }

The to_vec() is a fresh heap allocation, needed because walk_up_resolve_reference takes &mut self and would otherwise alias the borrow on the underlying Vec<(Ident, ReferenceId)>.

Rewrite in-place with a retain-style write cursor — each (Ident, ReferenceId) is read by value (both are Copy), which detaches it from the borrow on the inner Vec. Unresolved refs are compacted forward; resolved refs are dropped via the final truncate. No temporary Vec, no allocation, no unsafe — safe bounds-checked indexing via new get / set methods on UnresolvedReferences.

let mut write_idx = checkpoint;
for read_idx in checkpoint..end {
    let (name, reference_id) = self.unresolved_references.get(read_idx);
    if !self.walk_up_resolve_reference(name, reference_id) {
        if write_idx != read_idx {
            self.unresolved_references.set(write_idx, name, reference_id);
        }
        write_idx += 1;
    }
}
self.unresolved_references.truncate(write_idx);

Allocation impact

cargo allocs, allocs_semantic.snap (baseline = parent commit, after var-hoist landed):

File Size Sys allocs before Sys allocs after Δ
checker.ts 2.92 MB 2,309 53 −2,256 (−98%)
App.tsx 415 kB 174 35 −139 (−80%)
binder.ts 193 kB 196 19 −177 (−90%)
kitchen-sink.tsx 733 kB 1,320 290 −1,030 (−78%)
pdf.mjs 567 kB 1,366 1,355 −11
antd.js 6.69 MB 1,072 1,072 0 (ES5 bundled code rarely closes over outer scope from function params)
RadixUI.jsx 2.5 kB 10 10 0

How I found this

After #22580 and #22590 the remaining semantic sys allocs were unexplained. Ran a backtrace-capturing System allocator wrapper on checker.ts and aggregated allocations by call stack — the overwhelming majority of captured allocations converged on a single site:

oxc_semantic::builder::SemanticBuilder::resolve_references_for_current_scope
  oxc_semantic::builder::SemanticBuilder as ...::Visit::visit_function
  ...

Tracing back to the slice_from(checkpoint).to_vec() line and confirming that the entire allocation pattern disappears with the in-place rewrite.

Why this matters for downstream consumers

resolve_references_for_current_scope fires once per function / arrow / catch — for normal TS / TSX code that's hundreds to thousands of calls per build. Each call was a fresh heap allocation; eliminating them removes that allocator pressure entirely.

For rolldown's preprocessing pipeline (which builds semantic 3-4 times per file across hundreds of files per bundle), the per-build savings compound.

No behavior change

The in-place algorithm preserves the original semantics exactly: unresolved_references[..checkpoint] is untouched, and [checkpoint..] is the list of refs that didn't resolve, in their original order. Same as the old code.

Test Plan

  • cargo test -p oxc_semantic --lib --tests — pass
  • cargo allocs — semantic snapshot updated to reflect the reductions

AI disclosure: drafted with Claude Code, reviewed manually.

@github-actions github-actions Bot added the A-semantic Area - Semantic label May 19, 2026
@codspeed-hq

codspeed-hq Bot commented May 19, 2026

Copy link
Copy Markdown

Merging this PR will not alter performance

✅ 57 untouched benchmarks
⏩ 3 skipped benchmarks1


Comparing claude/perf-semantic-resolve-refs-no-tempvec (d6faed5) with claude/perf-semantic-var-hoist-smallvec (e8731d7)2

Open in CodSpeed

Footnotes

  1. 3 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports.

  2. No successful run was found on claude/perf-semantic-var-hoist-smallvec (7abee9b) during the generation of this report, so 81f9954 was used instead as the comparison base. There might be some changes unrelated to this pull request in this report.

@Dunqing Dunqing marked this pull request as draft May 19, 2026 15:55
@Dunqing Dunqing changed the base branch from main to graphite-base/22599 May 20, 2026 01:30
@Dunqing Dunqing force-pushed the claude/perf-semantic-resolve-refs-no-tempvec branch from 767e575 to 3a7691e Compare May 20, 2026 01:30
@Dunqing Dunqing changed the base branch from graphite-base/22599 to claude/perf-semantic-var-hoist-smallvec May 20, 2026 01:30

Dunqing commented May 20, 2026

Copy link
Copy Markdown
Member Author

How to use the Graphite Merge Queue

Add either label to this PR to merge it via the merge queue:

  • 0-merge - adds this PR to the back of the merge queue
  • hotfix - for urgent changes, fast-track this PR to the front of the merge queue

You must have a Graphite account in order to use the merge queue. Sign up using this link.

An organization admin has enabled the Graphite Merge Queue in this repository.

Please do not merge from GitHub as this will restart CI on PRs being processed by the merge queue.

This stack of pull requests is managed by Graphite. Learn more about stacking.

@Dunqing Dunqing changed the base branch from claude/perf-semantic-var-hoist-smallvec to graphite-base/22599 May 20, 2026 01:40
@Dunqing Dunqing force-pushed the graphite-base/22599 branch from 81d2fee to dc53600 Compare May 20, 2026 01:42
@Dunqing Dunqing force-pushed the claude/perf-semantic-resolve-refs-no-tempvec branch from 3a7691e to 4236168 Compare May 20, 2026 01:42
@Dunqing Dunqing changed the base branch from graphite-base/22599 to claude/perf-semantic-var-hoist-smallvec May 20, 2026 01:42
@Dunqing Dunqing changed the base branch from claude/perf-semantic-var-hoist-smallvec to graphite-base/22599 May 20, 2026 02:50
@Dunqing Dunqing force-pushed the claude/perf-semantic-resolve-refs-no-tempvec branch from 4236168 to f1886cc Compare May 20, 2026 02:51
@Dunqing Dunqing force-pushed the graphite-base/22599 branch from dc53600 to 223eaaf Compare May 20, 2026 02:51
@Dunqing Dunqing changed the base branch from graphite-base/22599 to claude/perf-semantic-var-hoist-smallvec May 20, 2026 02:51
@Dunqing Dunqing changed the base branch from claude/perf-semantic-var-hoist-smallvec to graphite-base/22599 May 20, 2026 09:37
@Dunqing Dunqing force-pushed the graphite-base/22599 branch from 223eaaf to fc5257e Compare May 20, 2026 10:16
@Dunqing Dunqing force-pushed the claude/perf-semantic-resolve-refs-no-tempvec branch from f1886cc to eb76d6d Compare May 20, 2026 10:16
@Dunqing Dunqing changed the base branch from graphite-base/22599 to claude/perf-semantic-var-hoist-smallvec May 20, 2026 10:16
@Dunqing Dunqing changed the base branch from claude/perf-semantic-var-hoist-smallvec to graphite-base/22599 May 20, 2026 10:21
@Dunqing Dunqing force-pushed the graphite-base/22599 branch from fc5257e to f0f2f65 Compare May 20, 2026 10:22
@Dunqing Dunqing force-pushed the claude/perf-semantic-resolve-refs-no-tempvec branch from eb76d6d to 929db20 Compare May 20, 2026 10:22
@Dunqing Dunqing changed the base branch from graphite-base/22599 to claude/perf-semantic-var-hoist-smallvec May 20, 2026 10:22
@Dunqing Dunqing marked this pull request as ready for review May 20, 2026 10:27
@Dunqing Dunqing force-pushed the claude/perf-semantic-resolve-refs-no-tempvec branch 3 times, most recently from ee83fb5 to 4439904 Compare May 20, 2026 13:07
@Dunqing Dunqing force-pushed the claude/perf-semantic-resolve-refs-no-tempvec branch from 4439904 to 494d5dc Compare May 20, 2026 13:11
@Dunqing Dunqing force-pushed the claude/perf-semantic-var-hoist-smallvec branch from f0f2f65 to e8731d7 Compare May 20, 2026 15:20
@Dunqing Dunqing force-pushed the claude/perf-semantic-resolve-refs-no-tempvec branch from 494d5dc to d515956 Compare May 20, 2026 15:21
graphite-app Bot pushed a commit that referenced this pull request May 20, 2026
## Summary

Adds `kitchen-sink.tsx` — a comprehensive synthetic TypeScript+JSX fixture maintained at [oxc-project/benchmark-files](https://github.com/oxc-project/benchmark-files) — to both `TestFiles::minimal()` (bench input set) and `TestFiles::complicated()` (alloc-tracking input set). The existing files in each set are untouched; this is a strict append.

## Why

The existing bench input set didn't reliably surface general-purpose perf wins above the ~1-2% measurement noise floor:
- #22580 (semantic pre-reserve) — visible because `binder.ts` exercises it
- #22594 (formatter buffer) — visible
- #22596 (minifier `try_fold_concat`) — **not visible** on the old set
- #22599 (semantic resolve-refs no-temp-Vec) — **not visible**
- #22603 (semantic var-hoist SmallVec) — **not visible**

The kitchen-sink fixes that by exercising every AST node, every transformer plugin, every minifier optimization opportunity, and every semantic step in one large file. Verified by re-benching #22596 against this fixture: **minifier mean −1.5%, min −3.7%** — above noise, signal confirmed.

## Fixture stats (cross-checked locally)

| Metric | Value |
|---|---|
| Source size | 21,117 lines / 732.90 kB |
| AST nodes | ~133,000 |
| Scopes | ~4,750 |
| Symbols | ~7,000 |
| Resolved references | ~16,000 |
| Semantic diagnostics | 0 errors / 0 warnings |

## Snap baselines

`tasks/track_memory_allocations/allocs_*.snap` updated with the kitchen-sink row across all 5 pipelines (parser / semantic / transformer / minifier / formatter). Future PRs that change allocation behavior on this fixture will produce a snap diff in CI.

## Bench-cleaner fix

`tasks/benchmark/benches/lexer.rs`'s `SourceCleaner` was missing `visit_ts_template_literal_type` — TypeScript type-level template literals (e.g. `` `${T}-${U}` `` in conditional / mapped types) are syntactically identical to value-level template literals, so the bench-mode lexer (without parser context) cannot distinguish them. Without the cleaner converting them to plain strings, kitchen-sink's type-level templates caused the lexer bench to swallow ~1 KB spans as a single `TemplateHead` and produce spurious "Unterminated string" / "Invalid Unicode escape" errors. One-line fix to mirror the existing `visit_template_literal` handling.

AI disclosure: drafted with Claude Code, reviewed manually.
@graphite-app graphite-app Bot force-pushed the claude/perf-semantic-var-hoist-smallvec branch 2 times, most recently from 7aba78a to 7abee9b Compare May 20, 2026 16:03
@graphite-app graphite-app Bot force-pushed the claude/perf-semantic-resolve-refs-no-tempvec branch from d515956 to d6faed5 Compare May 20, 2026 16:03
@graphite-app graphite-app Bot added the 0-merge Merge with Graphite Merge Queue label May 21, 2026
@camc314 camc314 self-assigned this May 21, 2026
@graphite-app

graphite-app Bot commented May 21, 2026

Copy link
Copy Markdown
Contributor

Merge activity

…ec (#22599)

## Summary

`SemanticBuilder::resolve_references_for_current_scope` is the early-resolution hook fired at function / arrow / catch boundaries — it resolves the references collected since a checkpoint against the outer scope chain so that e.g. `function f(x = outerRef) {}` correctly binds `outerRef` to the outer scope rather than the function body.

The old implementation copied the pending slice into a fresh `Vec` on every call:

```rust
let refs = self.unresolved_references.slice_from(checkpoint).to_vec();
self.unresolved_references.truncate(checkpoint);
for (name, reference_id) in refs { ... }
```

The `to_vec()` is a fresh heap allocation, needed because `walk_up_resolve_reference` takes `&mut self` and would otherwise alias the borrow on the underlying `Vec<(Ident, ReferenceId)>`.

Rewrite in-place with a retain-style write cursor — each `(Ident, ReferenceId)` is read by **value** (both are `Copy`), which detaches it from the borrow on the inner Vec. Unresolved refs are compacted forward; resolved refs are dropped via the final `truncate`. No temporary `Vec`, no allocation, **no `unsafe`** — safe bounds-checked indexing via new `get` / `set` methods on `UnresolvedReferences`.

```rust
let mut write_idx = checkpoint;
for read_idx in checkpoint..end {
    let (name, reference_id) = self.unresolved_references.get(read_idx);
    if !self.walk_up_resolve_reference(name, reference_id) {
        if write_idx != read_idx {
            self.unresolved_references.set(write_idx, name, reference_id);
        }
        write_idx += 1;
    }
}
self.unresolved_references.truncate(write_idx);
```

## Allocation impact

`cargo allocs`, `allocs_semantic.snap` (baseline = parent commit, after var-hoist landed):

| File | Size | Sys allocs before | Sys allocs after | Δ |
|---|---|---|---|---|
| `checker.ts` | 2.92 MB | 2,309 | **53** | **−2,256 (−98%)** |
| `App.tsx` | 415 kB | 174 | **35** | **−139 (−80%)** |
| `binder.ts` | 193 kB | 196 | **19** | **−177 (−90%)** |
| `kitchen-sink.tsx` | 733 kB | 1,320 | **290** | **−1,030 (−78%)** |
| `pdf.mjs` | 567 kB | 1,366 | 1,355 | −11 |
| `antd.js` | 6.69 MB | 1,072 | 1,072 | 0 (ES5 bundled code rarely closes over outer scope from function params) |
| `RadixUI.jsx` | 2.5 kB | 10 | 10 | 0 |

## How I found this

After [#22580](#22580) and [#22590](#22590) the remaining semantic sys allocs were unexplained. Ran a backtrace-capturing `System` allocator wrapper on `checker.ts` and aggregated allocations by call stack — the overwhelming majority of captured allocations converged on a single site:

```
oxc_semantic::builder::SemanticBuilder::resolve_references_for_current_scope
  oxc_semantic::builder::SemanticBuilder as ...::Visit::visit_function
  ...
```

Tracing back to the `slice_from(checkpoint).to_vec()` line and confirming that the entire allocation pattern disappears with the in-place rewrite.

## Why this matters for downstream consumers

`resolve_references_for_current_scope` fires once per function / arrow / catch — for normal TS / TSX code that's hundreds to thousands of calls per build. Each call was a fresh heap allocation; eliminating them removes that allocator pressure entirely.

For rolldown's preprocessing pipeline (which builds semantic 3-4 times per file across hundreds of files per bundle), the per-build savings compound.

## No behavior change

The in-place algorithm preserves the original semantics exactly: `unresolved_references[..checkpoint]` is untouched, and `[checkpoint..]` is the list of refs that didn't resolve, in their original order. Same as the old code.

## Test Plan

- [x] `cargo test -p oxc_semantic --lib --tests` — pass
- [x] `cargo allocs` — semantic snapshot updated to reflect the reductions

AI disclosure: drafted with Claude Code, reviewed manually.
@graphite-app graphite-app Bot force-pushed the claude/perf-semantic-var-hoist-smallvec branch from 223ae1a to e862c15 Compare May 21, 2026 07:56
@graphite-app graphite-app Bot force-pushed the claude/perf-semantic-resolve-refs-no-tempvec branch from d6faed5 to 4f289f1 Compare May 21, 2026 07:57
Base automatically changed from claude/perf-semantic-var-hoist-smallvec to main May 21, 2026 08:01
@graphite-app graphite-app Bot merged commit 4f289f1 into main May 21, 2026
31 checks passed
@graphite-app graphite-app Bot removed the 0-merge Merge with Graphite Merge Queue label May 21, 2026
@graphite-app graphite-app Bot deleted the claude/perf-semantic-resolve-refs-no-tempvec branch May 21, 2026 08:02
Dunqing added a commit that referenced this pull request May 26, 2026
### 🚀 Features

- e857b0c napi/minify: Expose legalComments option and result (#20370)
(Boshen)
- 661132d parser: More friendly error messages for rest assignment
target and rest binding element (#22719) (sapphi-red)
- ee659b6 transformer/legacy-decorator: Add `strictNullChecks` option
for nullable-union design:type (#22266) (Kyle Cannon)

### 🐛 Bug Fixes

- e1d064e transformer/class-properties: Reparent lifted private method
helpers (#22716) (Cameron)
- 4ac0fca minifier: Preserve `0 && (module.exports = { ... })`
cjs-module-lexer hint (#22729) (Dunqing)
- 40ff611 minifier: Mark peephole loop changed when dropping
dead-after-throw statement (#22722) (Dunqing)
- 2f7b210 codegen: Emit pife-arrow/function leading comments inside the
wrap (#22720) (Dunqing)
- e184f74 parser: Improve invalid `import` property access diagnostic
(#22693) (camc314)
- 7baed9c transformer/private-method: Clear inherited strict flags
(#22508) (camc314)
- a9ad27e parser: Keep annotation comments leading without preceding
newline (#22711) (Dunqing)
- 9ea4d64 minifier: Re-evaluate pure/no-side-effects flags after
peephole inlining (#22595) (Dunqing)
- 07afbb6 minifier: Drop empty-body IIFE wrapper when called with
arguments (#22589) (Dunqing)
- fa7c463 semantic: Correct TS enum member symbol spans (#22689)
(camc314)
- 26b9396 semantic: Resolve parameter decorators outside parameter scope
(#22623) (camc314)
- b284045 parser: Switch to module goal eagerly on `export` (#22684)
(Boshen)
- dfa931d semantic: Propagate unresolved auto-increment enum value
instead of defaulting to 0 (#22646) (Dunqing)
- 69a6ba6 transformer/legacy-decorator: Emit Array for ReadonlyArray<T>
in decorator metadata (#22265) (Kyle Cannon)
- e421ef0 transformer/legacy-decorator: Return runtime binding for
design:type (#22640) (Dunqing)
- d61e1d7 codegen: Preserve verbatim text of pure/no-side-effects
comments (#22525) (Dunqing)
- 702b14e minifier: Preserve IIFE structure in DCE-only mode (#22547)
(Dunqing)
- 917da24 parser: Apply PURE comment through member-access chains
(#22566) (Dunqing)
- a069b1c codegen: Preserve quotes for cjs-module-lexer equality strings
(#22551) (Dunqing)

### ⚡ Performance

- 2f623b0 semantic: Skip unresolved checks for re-exports (#22660)
(camc314)
- 0d9553d semantic: Early-exit `check_object_expression` for objects
with <2 properties (#22668) (Dunqing)
- d721ad9 semantic: Use direct grandparent lookup for TS type parameters
(#22658) (camc314)
- 0aff288 semantic: Reorder numeric literal strict mode checks (#22657)
(camc314)
- 4d5ddb1 semantic: Reorder binding identifier checks (#22656) (camc314)
- e32acd8 semantic: Reorder identifier ambient binding check (#22653)
(camc314)
- 09fe178 semantic: Reorder ident reference strict mode check (#22652)
(camc314)
- 4b6add2 semantic: Avoid duplicate ident clone for bindings (#22663)
(camc314)
- 82f9662 parser: Check identifier kind before context flag (#22662)
(camc314)
- d7cd951 parser: Fast path identifier parsing and inline operator
helpers (#22650) (Boshen)
- 7b84314 semantic: Use direct byte access for numeric leading-zero
check (#22642) (camc314)
- 0345a31 semantic: Pre-size class elements hash map (#22618) (camc314)
- 04d3065 minifier: Drop per-call buffers in try_fold_concat (#22596)
(Dunqing)
- 4f289f1 semantic: Resolve_references_for_current_scope without a temp
Vec (#22599) (Dunqing)
- e862c15 semantic: Avoid heap alloc for var hoist scope ids (#22603)
(Dunqing)
- 8ff8674 semantic: Early return if `excess` is `0` in
`Stats::increase_by` (#22616) (camc314)
- 7a4120e semantic: Pre-reserve unresolved_references using
Stats::references (#22580) (Dunqing)

Co-authored-by: Dunqing <29533304+Dunqing@users.noreply.github.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

A-semantic Area - Semantic

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants